Add SchemaManager::waitForSearchIndexes()

[GromNaN]

Q	A
Type	feature
BC Break	no
Fixed issues	PHPORM-399

Summary

This method waits until all search indexes for specified (or all) document classes are both queryable and have indexed all documents in their corresponding collections.

• It takes up to three arguments: a list of document class names ($classNames, optional), a maximum wait time in milliseconds ($maxTimeMs, default 10,000 ms), and a delay between checks in milliseconds ($waitTimeMs, default 100 ms).
• The method validates input arguments, then iterates over the document classes to identify the search indexes that need to be checked.
• In a loop, it checks whether all required search indexes exist and are marked as queryable.
• For queryable indexes, it further verifies if all documents have been indexed by comparing the count of indexed documents with the total collection count, using aggregation pipelines appropriate for either 'search' or 'vectorSearch' index types. (removed because unreliable)
• If not all indexes are ready, it waits for $waitTimeMs before checking again, until either all indexes are ready or the maximum wait time ($maxTimeMs) is exceeded (in which case an exception is thrown).

This addition provides a mechanism for applications to block until search indexes are fully prepared for querying, ensuring consistency and readiness before proceeding with operations that depend on those indexes.

Similar implementation: https://github.com/mongodb-labs/pymongo-search-utils/blob/dddd51f6e1784caa43e0dc66ec8d9487e0762d58/pymongo_search_utils/index.py#L259-L298

[GromNaN]

I reviewed the full implementation to remove the counting of collection vs search index. In fact when documents are inserted with wc:majority before creating the search index, then the status of the search index is queryable: true only when all documents are synchronized. (at least from my tests with atlas local container).

[jmikola]

FWIW, I asked about this in #atlas-search as I couldn't find this discussed in the Atlas Search docs. In any event, no action needed on your part here.

[jmikola]

What a marvelous function array_column() is. Thank you, @ramsey.